Skew-t partially linear mixed-effects models for AIDS clinical studies

Research output: Contribution to journalArticle

Abstract

We propose partially linear mixed-effects models with asymmetry and missingness to investigate the relationship between two biomarkers in clinical studies. The proposed models take into account irregular time effects commonly observed in clinical studies under a semiparametric model framework. In addition, commonly assumed symmetric distributions for model errors are substituted by asymmetric distribution to account for skewness. Further, informative missing data mechanism is accounted for. A Bayesian approach is developed to perform parameter estimation simultaneously. The proposed model and method are applied to an AIDS dataset and comparisons with alternative models are performed.

Original languageEnglish
Pages (from-to)899-911
Number of pages13
JournalJournal of Biopharmaceutical Statistics
Volume26
Issue number5
DOIs
StatePublished - Sep 2 2016

Keywords

  • Bayesian inference
  • longitudinal data
  • missing data
  • partially linear mixed-effects models
  • skew distribution

Fingerprint Dive into the research topics of 'Skew-t partially linear mixed-effects models for AIDS clinical studies'. Together they form a unique fingerprint.

  • Cite this